medium term pasa process description - aemo

Australian Energy Market Operator Ltd ABN 94 072 010 327 www.aemo.com.au [email protected]

NEW SOUTH WALES QUEENSLAND SOUTH AUSTRALIA VICTORIA AUSTRALIAN CAPITAL TERRITORY TASMANIA WESTERN AUSTRALIA

MEDIUM TERM PASA PROCESS DESCRIPTION

PREPARED BY: AEMO Forecasting

DOCUMENT REF: 42

VERSION: 6.1

STATUS: FINAL

Approved for distribution and use by:

APPROVED BY: Nicola Falcon

TITLE: Group Manager – Forecasting

DATE: 7 / 9 / 20

http://www.aemo.com.au/

mailto:[email protected]


Australian Energy Market Operator Ltd ABN 94 072 010 327 www.aemo.com.au [email protected]

NEW SOUTH WALES QUEENSLAND SOUTH AUSTRALIA VICTORIA AUSTRALIAN CAPITAL TERRITORY TASMANIA WESTERN AUSTRALIA

VERSION RELEASE HISTORY

Version Effective Date Summary of Changes

1.0 27/4/2006 SOPP

2.0 22/3/2013 Systems Capability

3.0 30/5/2013 Systems Capability

4.0 25/11/2016 Forecasting & Planning

4.1 08/6/2017 Supply Planning

5.0 15/8/2017 Supply Planning

5.1 7/5/2017 Forecasting



http://www.aemo.com.au/

mailto:[email protected]


Doc Ref: 42 of 40

CONTENTS

1. INTRODUCTION 5

2. MT PASA PROCESS AND RULES REQUIREMENTS 7

3. MT PASA INPUTS 8

3.1. Market participant inputs 8

3.2. AEMO inputs 9

4. MT PASA SOLUTION PROCESS 13

4.1. NEM Representation 13

4.2. Overview of Modelling Approach 13

4.3. MT PASA Reliability Run 13

4.4. MT PASA Loss of Load Probability (LOLP) Run 15

4.5. Comparison of Model Features 16

5. MT PASA OUTPUTS 18

APPENDIX A. MT PASA PROCESS ARCHITECTURE 20

APPENDIX B. MEDIUM TERM DEMAND FORECASTING PROCESS 21

B.1 Reliability Run Demand Traces 21

B.2 Loss of Load Probability Run Demand Traces 22

B.3 MT PASA Daily maximum and minimum Demand Values 22

APPENDIX C: PAIN SHARING 27

APPENDIX D: CALCULATION OF TRANSFER LIMITS 28

APPENDIX E: GRAPHICAL OUTPUTS 29

APPENDIX F: MT PASA OUTPUT TABLES 33

APPENDIX G: “PLAIN ENGLISH” REPORT ON CONSTRAINTS 40

TABLES

Table 1 Rules requirements ..................................................................................................................................................... 7

Table 2 AEMO Demand Definitions .................................................................................................................................... 10

Table 3 Comparison of MT PASA run features ............................................................................................................... 16

Table 4 MT PASA Outputs Specified in NER 3.7.2(f)(6) produced by Reliability Run ........................................ 18

Table 5 Example: Maximum dates and time for Ex VRE Demand in February.................................................... 22

FIGURES

Figure 1 MT PASA Reliability Run case construction ...................................................................................................... 14

Figure 2 MT PASA Data Flows ............................................................................................................................................... 20


Doc Ref: 42 of 40

Figure 3 AEMO Operational Demand Diagram ............................................................................................................... 21

Figure 4 Native demand components ................................................................................................................................ 23

Figure 5 Method for developing reported MT PASA daily demand forecasts ..................................................... 24

Figure 6 Development of weekly factor profile ............................................................................................................... 25

Figure 7 Development of weekday factor profile ........................................................................................................... 25

Figure 8 Example of daily demand calculations .............................................................................................................. 26

Figure 9 Assessment of Reliability Standard ..................................................................................................................... 29

Figure 10 Annual distribution of Unserved energy (User to select region and year) ........................................... 30

Figure 11 Size of Unserved Energy events by month (User to select POE demand level, region and year) 30

Figure 12 Severity and Frequency of Unserved Energy (User to select region and year) .................................... 31

Figure 13 Interconnector flow limits (User to select interconnector) .......................................................................... 31

Figure 14 Supply demand breakdown and maintenance period overview from LOLP run (User to select

region and year) ........................................................................................................................................................ 32

Figure 15 Monthly expected unserved energy (User to select region and year) ................................................... 32

Figure 16 Example constraints viewer ................................................................................................................................... 40


Doc Ref: 42 of 40

1. INTRODUCTION

The National Electricity Rules (the Rules) clause 3.7.1 require the Australian Energy Market Operator

(AEMO) to administer the projected assessment of system adequacy (PASA) processes.

The PASA is the principal method for indicating to the National Electricity Market (NEM) the

forecast adequacy of power system security and supply reliability over the next 24 months. The

Rules require AEMO to administer the PASA over two timeframes:

1. Medium Term PASA (MT PASA): this assessment covers the 24 month period starting from the

first Sunday after publication. It is updated and published weekly to a daily resolution.

2. Short Term PASA (ST PASA): this assessment covers the six trading days starting from the end of

the trading day covered by the most recently published pre-dispatch schedule. It is updated and

published every two hours to a trading interval resolution.

MT PASA assesses power system security and reliability under a minimum of 10% Probability of

Exceedance (POE) and 50% POE demand conditions based on generator availabilities submitted

by market participants, with due consideration to planned transmission and relevant distribution

outages and limits1. The reliability standard is a measure of the effectiveness, or sufficiency, of

installed capacity to meet demand and is defined in clause 3.9.3C of the Rules.

The MT PASA process includes (but is not limited to):

• Information collection from Scheduled Generators, Market Customers, Transmission Network

Service Providers and Market Network Service Providers about their intentions (as appropriate)

for:

− Generation, transmission and market network service maintenance scheduling.

− Intended plant availabilities.

− Energy constraints.

− Other plant conditions which could materially impact upon power system security and the

reliability of supply.

− Significant changes to load forecasts.

• Analysis of medium-term power system security and reliability of supply.

• Forecasts of supply and demand.

• Provision of information that allows participants to make decisions about supply, demand and

outages of transmission networks for the next 24 months2.

• Publication of sufficient information to allow the market to operate effectively with a minimal

amount of intervention by AEMO.

The MT PASA process is administered according to the timeline set out in the Spot Market

Operations Timetable3 (timetable) in accordance with the Rules.

This document fulfils AEMO’s obligation under clause 3.7.2(h) of the Rules to document the

procedure used in administering the MT PASA.

1 Constraints will be invoked on embedded generators connected to the DNSP network when there is an impact on TNSP equipment.

When there is no impact on the TNSP network, constraints will not be applied. DNSPs should coordinate with generators and the

generators should reflect the MW availability accordingly. For further information see https://www.aemo.com.au/-/media/Files/

Electricity/NEM/Security_and_Reliability/Power_System_Ops/Procedures/SO_OP_3718---Outage-Assessment.pdf 2 The information on generating unit availabilities and daily demands is published in the Region Availability report for the next 36

months. 3 http://www.aemo.com.au/-/media/Files/Electricity/NEM/Security_and_Reliability/Dispatch/Spot-Market-Operations-Timetable.pdf

https://www.aemo.com.au/-/media/‌Files/‌Electricity/‌NEM/Security_and_Reliability/Power_System_Ops/Procedures/SO_OP_3718---Outage-Assessment.pdf

https://www.aemo.com.au/-/media/‌Files/‌Electricity/‌NEM/Security_and_Reliability/Power_System_Ops/Procedures/SO_OP_3718---Outage-Assessment.pdf

http://www.aemo.com.au/-/media/Files/Electricity/NEM/Security_and_Reliability/Dispatch/Spot-Market-Operations-Timetable.pdf


Doc Ref: 42 of 40

1.1.1. Glossary

Terms defined in the National Electricity Law and the Rules have the same meanings in these

Procedures unless otherwise specified in this clause.

Defined terms/Terms defined in the Rules are intended to be identified in these Procedures by

italicising them, but failure to italicise a defined term does not affect its meaning.

The words, phrases and abbreviations in the table below have the meanings set out opposite them

when used in these Procedures.

Term Definition

AEMO Australian Energy Market Operator

ASEFS Australian Solar Energy Forecasting System

AWEFS Australian Wind Energy Forecasting System

ESOO Electricity Statement of Opportunities

LP Linear Program

LRC Low Reserve Condition

MMS Electricity Market Management System

NEM National Electricity Market

Rules National Electricity Rules (the Rules)

PASA Projected Assessment of System Adequacy

ST PASA: Short term projected assessment of system adequacy

MT PASA: Medium term projected assessment of system adequacy

POE Probability of Exceedance

RHS Right Hand Side of a constraint equation

Timetable Spot Market Operations Timetable

UIGF Unconstrained Intermittent Generation Forecast

USE Unserved Energy

VRE Variable renewable energy

1.1.2. Interpretation

These Procedures are subject to the principles of interpretation set out in Schedule 2 of the

National Electricity Law.


Doc Ref: 42 of 40

2. MT PASA PROCESS AND RULES REQUIREMENTS

The PASA is a comprehensive program for collecting and analysing information to assess medium-

and short-term power system security and reliability of supply prospects. This is so that Registered

Participants are properly informed to enable them to make decisions about supply, demand and

outages of transmission networks for periods up to 36 months in advance. MT PASA assesses the

adequacy of expected electricity supply to meet demand across the two-year horizon through

regularly identifying and quantifying any projected failure to meet the reliability standard.

MT PASA incorporates two separate functions:

1. A high frequency three-hourly information service (the ‘three-hourly report’) that gives a

regional breakdown of the supply situation over a 36 month horizon, taking into account

participant submissions on availability (the REGIONAVAILABILITY report).

2. A weekly assessment of system reliability, including provision of information on demand, supply

and network conditions.

AEMO must review and publish the MT PASA outputs in accordance with the frequency specified in

clause 3.7.2(a), covering the period starting from the Sunday after day of publication with a daily

resolution. Additional updated versions of MT PASA may be published by AEMO in the event of

changes which, in the judgement of AEMO, are materially significant and should be communicated

to Registered Participants.

Each party’s responsibilities in preparing MT PASA (summarised in Table 1 below) are also defined

in this clause.

Table 1 Rules requirements

Responsible

Party

Action Rules Requirement

AEMO Prepare the following MT PASA inputs:

• Forecasts of the 10% probability of exceedence daily peak load

and the most probable daily peak load

• Network constraints forecasts

• Unconstrained intermittent generation forecasts for semi-

scheduled generating unit

• The capabilities of generating units for which formal

commitments have been made for construction or installation

3.7.2(c)

Scheduled

Generator or

Market Participant

Submit to AEMO the following MT PASA inputs:

• PASA availability of each scheduled generating unit, scheduled

load or scheduled network service

• Weekly energy constraints applying to each scheduled generating

unit or scheduled load

3.7.2(d)

Network Service

Providers

Provide AEMO the following information:

• Outline of planned network outages

• Any other information on planned network outages that is

reasonably requested by AEMO

3.7.2(e)

AEMO Prepare and publish the MT PASA outputs 3.7.2(f)


Doc Ref: 42 of 40

3. MT PASA INPUTS

Inputs used in the MT PASA process are provided by AEMO and market participants. They are

discussed in detail below.

3.1. Market participant inputs

Market participants and Scheduled Generators are required to submit the following data in

accordance with the timetable, covering a 36 month period from the Sunday after the day of

publication of MT PASA.

3.1.1. Generating unit availabilities for MT PASA

• Generating unit PASA availabilities:

MT PASA uses PASA availabilities of generating units. PASA availability includes the generating

capacity in service as well as the generating capacity that can be delivered with 24 hours’

notice.

• As per clause 3.7.2(d)(1), Generators are required to provide the expected daily MW capacity of

each scheduled generating unit or scheduled load for the next 36 months and 24 months

respectively. The actual level of generation available at any particular time will depend on the

condition of the generating plant, which includes factors such as age, outages, and wear.

Another important factor with respect to output is the reduction in thermal efficiency with

increasing temperature.

• Generators should take into account the ambient weather conditions expected at the time

when the Region where the generating unit is located experiences the 10% Probability of

Exceedance (POE) peak load.

• Generating unit energy availabilities:

Generating plant such as hydroelectric power stations cannot generally operate at maximum

capacity indefinitely because their energy source may become exhausted. Gas and coal plants

can have energy constraints due to contracted fuel arrangements or emissions restrictions.

Under clause 3.7.2(d)(2), scheduled generating units with a weekly energy constraint (referred

to as energy constrained plant) are required to submit that weekly energy limit in MWh for all

relevant weeks over the upcoming 36-month period commencing from the first Sunday after

the latest MT PASA run.

AEMO may also use other information available such as that provided through the Generator

Energy Limitation Framework (GELF) or generator surveys to develop daily, monthly, annual

and/or biennial energy constraints for MT PASA modelling.

The energy limits should be determined by generators, taking into account:

− The potential for fuel stockpiles or water storages to fluctuate in the short term.

− The generator’s capability to replenish stockpiles and storages if depletion occurs.

• Wind turbine and large-scale solar availabilities:

To help AEMO fulfil its obligation under clause 3.7.2(c)(4), participants who operate such units

are required to submit local limit information on their wind turbine or solar availability to

AEMO. This information is used to augment historical generation data, to develop

unconstrained intermittent generation forecasts. Further details are provided in Section 3.2.1.


Doc Ref: 42 of 40

3.1.2. Network outages and Interconnector availabilities

Under clause 3.7.2(e), Network Service Providers must provide AEMO with an outline of planned

network outages and any other information on planned network outages reasonably requested by

AEMO. This includes interconnector availability information (e.g. Basslink). The planned network

outages are converted into network constraints by AEMO. This process is further discussed in

Section 3.2.3.

3.2. AEMO inputs

3.2.1. Plant availabilities for MT PASA

AEMO prepares other plant availability data, not provided by market participants:

• Semi-scheduled wind and solar generation forecasts:

AEMO is required to produce an unconstrained intermittent generation forecast (UIGF) for each

semi-scheduled generating unit for each day in accordance with clause 3.7.2(c)(4).

AEMO develops the UIGF using historically observed generation outputs for wind and solar

units for at least eight reference years. These outputs reflect the weather conditions that

underlie the demand traces for those reference years, ensuring that any correlation between

VRE generation and demand is preserved.

Where historical generation data is unavailable or unsuitable, AEMO may use historical

meteorological data for the site, and an energy conversion model based on the generator

technology to develop a generation forecast.

• Non-scheduled generation forecasts:

In accordance with clause 3.7.2(f)(2), AEMO is required to prepare and publish the aggregated

MW allowance (if any) to be made by AEMO for generation from non-scheduled generating

systems.

The non-scheduled generation profiles have two parts: large non-scheduled wind and solar

generation (refer to Table 2 for further details) and small non-scheduled generation. The large

non-scheduled wind and solar generation forecasts are calculated based on historically-

observed generation outputs over at least eight reference years, while the small non-scheduled

generation forecasts are consistent with figures published in AEMO’s demand forecasts4.

The small non-scheduled generation forecasts for units under 30MW are used as an input to

the MT PASA operational demand forecasting process and are not modelled explicitly.

• Demand Side Participation:

Demand Side Participation (DSP) includes all short-term reductions in demand in response to

temporary price increases (in the case of retailers and customers) or adverse network loading

conditions (in the case of networks). An organised, aggregated response may also be possible.

From the transmission network perspective, consumers may effectively reduce demand by

turning off electricity-using equipment or starting up on-site generators.

• Future generation:

Consistent with clause 3.7.2(c)(2), scheduled, semi-scheduled or large non-scheduled

generation projects with a commitment to construct or install5 are also modelled in MT PASA.

4 Available at http://forecasting.aemo.com.au/ 5 Information on the criteria used by AEMO to classify projects as committed can be found at https://aemo.com.au/en/energy-

systems/electricity/national-electricity-market-nem/nem-forecasting-and-planning/forecasting-and-planning-data/generation-

information

http://forecasting.aemo.com.au/

https://aemo.com.au/en/energy-systems/electricity/national-electricity-market-nem/nem-forecasting-and-planning/forecasting-and-planning-data/generation-information




Doc Ref: 42 of 40

Before the unit is registered, PASA availability for a committed scheduled generating unit is

estimated based on participant information regarding the commercial use date and seasonal

capacity. The Generator information page reports this information6.

The unit is entered into a Future Generation table that is referenced during modelling to include

all “committed but not registered” units. Once the unit is registered, it is removed from the Future

Generation table.

In the case of scheduled generators, the Generator that owns the unit is then responsible for

submitting MT PASA unit offer data to AEMO.

In the case of semi-scheduled generators and large non-scheduled generators, AEMO applies

availability traces for the unit for use in modelling, developed through either:

− Using a “shadow generator” based on existing VRE generation of a similar technology type

in close proximity; or

− Using meteorological data for the generation site, and assuming an energy conversion

model based on a similar technology type.

3.2.2. Demand forecasts

AEMO develops a range of demand forecasts for MT PASA that are used for both modelling and

reporting obligations. Table 2 shows the definitions of the different types of demand that are

referenced in this document.

For a more detailed explanation of the calculation of demand forecasts, please consult Appendix B.

Table 2 AEMO Demand Definitions

Demand Type Definition Description

Underlying Customer

consumption

Consumption on premises (“behind the meter”) including demand

supplied by rooftop PV and battery storage.

Delivered Underlying – PV –

battery

The energy the consumer (either residential or business) withdraws

from the electricity grid.

Native Delivered + (network

losses)

Total generation fed into the electricity grid. May be specified as

“sent-out” (auxiliary load excluded) or “as-generated” (auxiliary load

included).

Includes both transmission and distribution losses.

Operational “sent-

out” 7

Native – Small Non-

Scheduled (“as sent

out”)

Demand met by generation “as sent out” by scheduled / semi-

scheduled / large non-scheduled generators.

Operational “as

generated”

Operational “as sent

out” + auxiliary loads

Demand met by generation “as generated” by scheduled / semi-

scheduled / large non-scheduled generators including demand on

generator premises (auxiliary load).

VRE Variable renewable

energy

Demand met by semi-scheduled and large non-scheduled

generators excluding the impact of network constraints. This is a

non-standard demand definition used for LOLP modelling.

Operational “ex

VRE”

Operational “sent out”

- VRE

Demand met by scheduled generators. This is a non-standard

demand definition used for LOLP modelling.

6 http://aemo.com.au/Electricity/National-Electricity-Market-NEM/Planning-and-forecasting/Generation-information

7 For details on operational demand please refer to demand definitions here https://aemo.com.au/en/energy-systems/electricity/

national-electricity-market-nem/system-operations/dispatch-information

http://aemo.com.au/Electricity/National-Electricity-Market-NEM/Planning-and-forecasting/Generation-information

https://aemo.com.au/en/energy-systems/electricity/national-electricity-market-nem/system-operations/dispatch-information

https://aemo.com.au/en/energy-systems/electricity/national-electricity-market-nem/system-operations/dispatch-information


Doc Ref: 42 of 40

Demand Type Definition Description

Non-scheduled Large + Small Non-

Scheduled

Demand met by large and small non-scheduled generators.

Large Non-

scheduled

Also referred to as

Significant Non-

Scheduled

Large non-scheduled generators include:

• Wind or solar generators >=30 MW

• Generators classified as non-scheduled but treated as scheduled

generators in dispatch.

MT PASA demand forecasts are summarised and the specific demand requirements for each of the

two modelling runs are discussed in further detail below.

The daily demand forecasts published in the REGION AVAILABILITY table are provided for

reporting purposes only and are not the demand traces tht are used in any of the MT PASA

reliability assessments.

MT PASA Modelling:

• Annual operational “sent-out” demand profiles, consisting of half-hourly demand values, with

energy consumption and maximum demand aligned with AEMO’s latest sent-out forecasts.

(Reliability Run).

• Abstract operational demand and VRE generation forecasts constructed, based on the

evaluation of the years of historical observations. The traces represent conditions of high

demand levels occurring coincidentally with low VRE generation output and are abstract since

these conditions are assumed every day (LOLP Run).

MT PASA Reporting - Clause 3.7.2(f)(1) – (4):

• Daily peak 10% POE and 50% POE demand met by scheduled and semi-scheduled generators

(clause 3.7.2(f)(1) and (1A))8, non-scheduled allowance (clause 3.7.2(f)(2)), and native demand

(clause 3.7.2(f)(3)), aligned with AEMO’s latest forecasts.

• Weekly 50% POE energy consumption (clause 3.7.2(f)(4)).

Reliability Run

The annual operational “sent-out” demand profiles used in MT PASA modelling identify and

quantify any projected breach of the reliability standard. For this purpose, both maximum demand

and energy consumption are important to capture, and the profile is developed considering past

trends, day of the week and public holidays. Auxiliary load is calculated directly in the modelling,

based on assumed auxiliary load scaling factors for each generator.

The actual demand differs from forecast, mainly due to weather. Statistically, it can be assumed

that the forecast error follows a normal distribution. Accordingly, a forecast can be qualified by the

probability that actual demand will exceed forecast demand or POE:

• A 10% POE forecast indicates a 10% chance that actual demand will exceed the forecast value

over the relevant period (i.e. peak demand will be exceeded once in 10 years).

• A 50% POE forecast indicates a 50% chance that actual demand will exceed the forecast value

over the relevant period.

The timing and regional spread of these weather events also impacts on demand – hot weather in

a single region on a weekend will impact demand (and potentially reliability) differently than a heat

wave that has been building for days with impact felt across multiple regions.

8 Note, this is not the same as operational demand as it excludes both large and small non-scheduled generation.


Doc Ref: 42 of 40

To capture the impact of weather variations on demand, at least 16 different annual demand

profiles (corresponding to model cases discussed in Section 4.3) are developed for each region,

based on different historic weather patterns and POE annual peak demand forecasts. While this

captures a reasonable range of different weather-driven demand conditions, it unavoidably

requires assumptions to be made about precisely when the annual peak demand could occur,

based on historical demand patterns, even though it is impossible to predict when the annual peak

demand will occur in future.

Loss of Load Probability Run

Appropriate timing of maintenance scheduling can reduce the likelihood of unserved energy in

times of high demand. Consequently, it is important that AEMO also considers the loss of load

probability in each period of the modelling horizon, assuming weather conditions resulting in a

combination of high demand and low VRE generation were to occur in that specific period, to help

guide outage scheduling.

The LOLP demand and VRE generation modelling traces are based on high demand and low VRE

generation conditions observed over the different reference years, assessed on a month-by-month

basis for each day of the week. The traces can be classed as “abstract” since each day is considered

independently of the next, assuming close to monthly 10% POE weather conditions occurring each

day. Summing daily energy consumption will not produce realistic annual energy consumption

forecasts. Each region is considered independently but allows for support from adjacent regions

across interconnectors.

3.2.3. Power transfer capabilities used in MT PASA

For MT PASA, AEMO is required to forecast network constraints known to AEMO at the time, under

clause 3.7.2(c)(3).

Network constraints used in MT PASA represent technical limits on operating the power system.

These limits are expressed as a linear combination of generation and interconnectors, which are

constrained to be less than, equal to or greater than a certain limit.

Information to formulate network constraint equations is provided to AEMO by Transmission

Network Service Providers (TNSPs) via Network Outage Scheduler (NOS)9 and limit advice. The

process of producing network constraint equations is detailed in the Constraint Formulation

Guidelines10. Within AEMO’s market systems, constraint equations are marked as system normal if

they apply to all plant in service. To model network or plant outages in the power system, separate

outage constraint equations are formulated and applied with system normal constraint equations.

AEMO continues to update and refine network constraints through its ongoing modelling projects.

MT PASA uses the latest version of ST PASA formulation constraints as a starting base, with

additional customised network constraints associated with future planned network and generation

upgrades. AEMO constructs system normal and outage constraint equations for the MT PASA time

frame. MT PASA modelling is conducted with system normal and approved planned network

outage constraints applied.

See Appendix D for further information on the calculation of transfer capabilities.

9 http://nos.prod.nemnet.net.au/nos 10 http://www.aemo.com.au/Electricity/National-Electricity-Market-NEM/Security-and-reliability/Congestion-information

http://nos.prod.nemnet.net.au/nos

http://www.aemo.com.au/Electricity/National-Electricity-Market-NEM/Security-and-reliability/Congestion-information


Doc Ref: 42 of 40

4. MT PASA SOLUTION PROCESS

4.1. NEM Representation

The power system model used within the MT PASA simulation is similar to the model applied for

AEMO’s wholesale electricity market systems:

The salient features of the power system model are:

• Single regional reference node (RRN) within each market region at which all demand within the

region is deemed to apply.

• Generators connected to the regional reference node via a “hub and spoke” model. Static

transmission loss factors are used to refer price data from the generator connection point to

the RRN of the host region.

• Flow between market regions via interconnectors, which provide transport for energy between

regions. Losses for flows over interconnectors are modelled using a dynamic loss model.

• Modelling of thermal, stability and energy constraints to be achieved by overlaying constraint

equations onto the market-based model.

4.2. Overview of Modelling Approach

MT PASA assessment is carried out at least weekly using two different model runs:

1. Reliability Run – to identify and quantify potential reliability standard breaches, and assess

aggregate constrained and unconstrained capacity in each region, system performance and

network capability

2. Loss of Load Probability Run – to assess days most at risk of load shedding.

These two runs are discussed in more detail in the following sections.

4.3. MT PASA Reliability Run

The MT PASA Reliability Run implements the reliability standard by assessing the level of unserved

energy and evaluating the likelihood of reliability standard breaches through probabilistic

modelling. The Reliability Run is conducted weekly.


Doc Ref: 42 of 40

The MT PASA Reliability Run uses at least 100 Monte-Carlo simulations11 on a set of predefined

cases to assess variability in unserved energy outcomes (see Figure 1). Demand and VRE

generation supply assumptions vary for each case, driven by different historical weather

conditions. Within a case, the Monte-Carlo simulations vary with respect to unplanned generation

outages based on historical forced outage rates.

Figure 1 MT PASA Reliability Run case construction

In total, at least 1,000 simulations are conducted for each year of the reliability assessment horizon

and are weighted to form the final estimate of USE. AEMO is currently updating MT PASA

weightings to match ESOO12 using weightings of 30.4% for 10% POE, 39.2% for 50% POE and

30.4% for 90% POE, where 90% POE demand will not be subject to simulation to minimise costs

(unless USE is expected to be non-zero, and simulations will therefore improve accuracy).

The objective function associated with the simulation is:

• Minimise total generation cost plus hydro storage violation cost subject to:

− Supply/demand balance.

− Unit capacity limits observed.

− Unit/power station/portfolio energy limits observed.

− Network constraints observed.

− DSP bounds observed.

The Reliability Run is conducted in three phases:

1. Generate random patterns of forced outages and determine any other stochastic parameters

required for each simulation run.

11 Probabilistic modelling involves many repetitions of the simulation model while applying random sampling to certain components

of the model. In MT PASA the random sampling is applied to the occurrence of forced outages for generation. Other uncertain

variables such as regional demand coincidence and VRE generation availability are varied through use of the different cases. 12 Expected implementation date is December 2020.

Final USE Weightings

Simulations

Weather Reference

Years

Demand

MT PASA Reliability

RunRun

10% Probability of

Exceedance

Demand

At least 58

Reference Years

At least 100

simulations with

random forced

outages

Simulations are

weighted at 30.4%

50% Probability of

Exceedance

Demand

At least 58

Reference Years

At least 100

simulations with

random forced

outages

Simulations are

weighted at 69.6%

90% Probability of

Exceedance

Demand

At least 8 Reference

Years

At least 100

simulations with

random forced

outages.

90% POE simulations

are only processed

when USE is expected

to be non-zero.


Doc Ref: 42 of 40

2. Split the two-year MT PASA horizon into two one-year periods that are solved at a reduced

level of time detail to allow long-term energy constraints to be optimised so that resources

subject to constraints are deployed at the most appropriate time. Inter-temporal constraints are

decomposed into a set of ending targets for each weekly time frame selected for use in phase

three.

3. Solve the entire horizon in shorter weekly steps with full half-hourly detail, using the weekly

allocation targets determined in phase two. MT PASA weekly energy limits are co-optimised

with dispatch of other resources, including VRE generation, to maximise the value of the energy

limited resource.

Most hydro generators are modelled with storages and their generation is subject to historically

assessed inflows and outflows from these storages. Annual energy limits are implemented through

the requirement that the storage at the end of the year must be equal to or greater than the

storage at the start of the year. Storage levels must also remain within upper and lower bounds.

During phase two, a series of optimal storage targets for each weekly period are set for use in

phase three. If these targets are not met in phase three, penalties are applied according to a series

of penalty bands that are low for small variations and high for large variations from target levels.

In addition to the storage targets, hydro generation is also constrained according to both PASA

availability and any MT PASA weekly energy constraints submitted. Weekly energy constraints for

all generation types are considered in both phase two and phase three, and cannot be violated.

Each simulation produces an estimate of annual USE, with the simulations providing insight into

the distribution of annual USE. AEMO uses a weighting of 39.2% for 50% POE and 30.4% for 10%

POE demand levels, to assess the expected USE as a weighted average across all simulations. The

90% POE demand levels are not normally modelled explicitly as AEMO assumes that USE will be

zero, which is reflected in the weightings provided.

If there are material levels of USE in 50% POE results, AEMO considers running additional demand

levels such as 90% POE. The USE outcomes in these simulations would then be weighted by 30.4%.

AEMO is developing a broader range of POE traces for modelling and will update this document

should any changes be made, including weightings. The expected annual USE value from the

simulations can be compared directly against the reliability standard. This allows AEMO to

accurately assess whether the reliability standard can be met. AEMO declares a LRC if the expected

value of USE across all simulations exceeds the reliability standard.

Pain sharing is not included. Instead, the annual USE reported in a region reflects the source of any

supply shortfall and is intended to provide participants with the most appropriate locational signals

to drive efficient market responses. (See Appendix C for a more detailed explanation).

4.4. MT PASA Loss of Load Probability (LOLP) Run

To determine days most at risk of load shedding, AEMO conducts a LOLP assessment for each day

in the two-year horizon, assuming that weather conditions associated with high demand and/or

low VRE generation availability were to occur on that day. The main objective is to determine

which days have higher relative risk of loss of load to help participants schedule outages outside of

these periods, and indicate when AEMO may be required to direct or contract for reserves under

the RERT.

The abstract operational demand and VRE generation traces discussed in Section 3.2.2 are used

for the LOLP run. A detailed explanation of trace construction is given in Appendix B.

The LOLP run uses a probabilistic modelling approach similar to the Reliability Run. Up to 500

simulations with random unplanned outages of scheduled generation are carried out. Energy

constraints are not included for LOLP modelling, as only one day at a time is modelled and there is


Doc Ref: 42 of 40

no optimisation over the full horizon. Network constraints incorporating system normal limits and

planned outages are used along with the MT PASA availability submitted by participants.

The loss of load probability is calculated by firstly determining the probability of loss of load in

each half hour of the day. For example, if 50 out of 500 simulations show loss of load, there is

approximately a ten percent chance of loss of load in that particular half hour. The maximum half-

hourly LOLP across all 48 half hours is reported as the LOLP for the day.

4.5. Comparison of Model Features

Table 3 shows the comparison of the key features of the two MT PASA modelling runs.

Table 3 Comparison of MT PASA run features

MT PASA Inputs

Property Reliability Run LOLP Run

Horizon 2 years

Frequency of Run Weekly

Simulations At least 100 per case Up to 500, one case only.

Resolution Half Hourly, returning a single half hour per day based on worst demand/supply conditions

Registration Using market system registration as a base including regions, interconnectors, generators,

transmission loss factors, interconnector loss models, fuel and regional reference node

memberships for generators

Demand At least eight half hourly demand traces for

each of 10% POE and 50% POE maximum

demand forecasts.

One half hourly abstract operational demand

trace based on the maximum operational “ex

VRE” demand observed in the half hourly

reference years

Generator

Capacity

As per participant MT PASA declarations

Generator Bid

Offers

SRMC calculated from heat rate, fuel price, VOM etc.

Generator

Forced/partial

outage modelling

Probabilistic assessment of forced outages over multiple simulations

Hydro Modelling Based on AEMO hydro storage model13 with

monthly inflows associated with average

levels of annual production.

Pumped storage modelled.

MT PASA Weekly energy constraints applied.

Energy limitations are not considered.

VRE (Semi

Scheduled)

Generation

At least eight historical weather traces,

correlated to demand traces

Traces based on extreme monthly demand

and VRE generation conditions observed in

the half hourly historical reference years

13 AEMO’s ‘Market modelling methodology report’ document contains details on the hydro storage model and can be found here:

https://www.aemo.com.au/energy-systems/electricity/national-electricity-market-nem/nem-forecasting-and-planning/scenarios-

inputs-assumptions-methodologies-and-guidelines

https://www.aemo.com.au/energy-systems/electricity/national-electricity-market-nem/nem-forecasting-and-planning/scenarios-inputs-assumptions-methodologies-and-guidelines

https://www.aemo.com.au/energy-systems/electricity/national-electricity-market-nem/nem-forecasting-and-planning/scenarios-inputs-assumptions-methodologies-and-guidelines


Doc Ref: 42 of 40

Non-scheduled

Generation

Large non-scheduled generation is modelled individually through traces.

Small non-scheduled generation (<30MW) is based on the most recent AEMO forecast which

can be found on the AEMO forecasting portal.14 Further details on the methodology for

forecasting non-scheduled demand can be found on the AEMO ESOO information page.15

Network

Representation

ST PASA formulation constraints with dynamic right hand side (RHS with network outages)

TNSP Limit Data Equipment ratings inclusive of seasonal variations required for evaluating generic constraint

RHS

Interconnector

forced outage

modelling

Not modelled

Demand Side

Participation

At least eight static Price/Quantity bands.

Rooftop PV Correlated to demand trace, but not explicitly modelled.

MT PASA Solution

Property Reliability Run LOLP Run

Purpose of run Assess level of unserved energy and the

likelihood of reliability standard breaches.

Assess the days at highest risk of loss of load

Type of run LP minimising total generation cost subject

to:

Supply = demand

Unit capacity limits observed

Generator Energy limits observed

Network constraints observed

Hydro storage bounds observed

LP minimising total generation cost subject to:

Supply = demand

Network constraints observed

Hydro storage bounds observed

MT PASA Outputs (See Appendix F for Detailed Description of Outputs)

Property Property Property

Low Reserve

Condition

Forecasts of low reserve conditions based on

expected annual USE

Unserved Energy Distribution of unserved energy on a half

hourly snapshot, daily, monthly and annual

basis.

Loss of Load

Probability

Highest half hourly LOLP on any given day.

Interconnector

Transfer

Capabilities

Interconnector transfer capabilities under

system normal conditions are published on

the AEMO website.16 Interconnector

capabilities in the presence of outages are

assessed during the Reliability Run.

14 Available at http://forecasting.aemo.com.au/ 15 See the Demand Forecasting Methodology information provided at https://aemo.com.au/energy-systems/electricity/national-

electricity-market-nem/nem-forecasting-and-planning/forecasting-and-reliability/nem-electricity-statement-of-opportunities-esoo 16 Published in the ‘Interconnector Capabilities report’ which can be found here https://www.aemo.com.au/energy-

systems/electricity/national-electricity-market-nem/system-operations/congestion-information-resource/network-status-and-

capability


https://aemo.com.au/energy-systems/electricity/national-electricity-market-nem/nem-forecasting-and-planning/forecasting-and-reliability/nem-electricity-statement-of-opportunities-esoo


https://www.aemo.com.au/energy-systems/electricity/national-electricity-market-nem/system-operations/congestion-information-resource/network-status-and-capability




Doc Ref: 42 of 40

Network

Constraint Impacts

When and where network constraints may

become binding on the dispatch of

generation or load

Projected

violations of Power

System Security

Reporting on any binding and violating

constraints that occur during modelling

5. MT PASA OUTPUTS

Under clause 3.7.2(f) of the Rules, AEMO must publish the MT PASA outputs as part of the MT

PASA process17. From a reliability perspective, the main MT PASA output is the forecast of any low

reserve condition and the estimated USE value.

The NER 4.8.4(a) defines an LRC as:

“Low reserve condition – when AEMO considers that the balance of generation capacity and

demand for the period being assessed does not meet the reliability standard as assessed in

accordance with the reliability standard implementation guidelines”.

Table 4 shows the MT PASA outputs produced by the Reliability Run. The outputs are based on

short-run marginal cost bidding rather than any estimate of strategic bidding to emulate observed

market behaviour. Given the probabilistic nature of the Reliability Run, distributions of simulated

outputs are reported in most instances.

Table 4 MT PASA Outputs Specified in NER 3.7.2(f)(6) produced by Reliability Run

MT PASA OUTPUT

SPECIFICATIONS NER 3.7.2(f)

MT PASA PUBLICATION OUTPUT DETAILS

(6) Identification and quantification

of:

(i) Any projected violations of power

system security

MT PASA Reliability Run Constraint solution outputs identifying

binding and violating constraints. If any

constraints are violated, it indicates that

there is a projected violation of power

system security.

(ii) Any projected failure to meet the

reliability standard assessed in

accordance with the RSIG

MT PASA Reliability Run Annual regional weighted average USE used

to identify LRC level if above the reliability

standard.

(iii) Deleted

(iv) Forecast interconnector transfer

capabilities and the discrepancy

between forecast interconnector

transfer capabilities and the forecast

capacity of the relevant

interconnector in the absence of

outages on the relevant

interconnector only

MT PASA Reliability Run

Constraint library & NOS

Interconnector Capability

Report

MT PASA Reliability Run will provide range

estimates of interconnector capabilities in

the presence of outages. The Interconnector

Capability Report will provide estimates of

interconnector capabilities under system

normal conditions. AEMO recommends

using the Constraint Library and the

Network Outage Schedule for accurate and

comprehensive information on applicable

constraints.

17 http://www.nemweb.com.au/REPORTS/CURRENT/MEDIUM_TERM_PASA_REPORTS/. A guide to the information contained in the MT

PASA is available in the form of a ‘MMS Data Model Report’ found here: https://www.aemo.com.au/energy-

systems/electricity/national-electricity-market-nem/data-nem/market-management-system-mms-data

http://www.nemweb.com.au/REPORTS/CURRENT/MEDIUM_TERM_PASA_REPORTS/


Doc Ref: 42 of 40

MT PASA OUTPUT

SPECIFICATIONS NER 3.7.2(f)

MT PASA PUBLICATION OUTPUT DETAILS

(iv) Forecast interconnector transfer

capabilities and the discrepancy

between forecast interconnector

transfer capabilities and the forecast

capacity of the relevant

interconnector in the absence of

outages on the relevant

interconnector only


Constraint library & NOS

MT PASA Reliability Run will provide range

estimates of interconnector capabilities in

the presence of outages. The Interconnector

Capability Report18 will provide estimates of

interconnector capabilities under system

normal conditions. AEMO recommends

using the Constraint Library and the

Network Outage Schedule for accurate and

comprehensive information on applicable

constraints.

(v) When and where network

constraints may become binding on

the dispatch of generation or load


Constraint Report

Constraints may bind at different times in

Reliability Run, depending on the demand

and VRE generation trace used, forced

outages and generation dispatch.

AEMO will also provide a “plain English”

report on constraints that provides further

details on generators impacted by binding

constraints19. Appendix G provides a link

with instructions for this report.

Appendix F shows a detailed list of output fields that will be published as part of the MT PASA

results sent to participants. Due to the high number of simulations and the quantity of data

produced during the runs, the results are aggregated before release to participants.

Where results are reported for a day on a half-hourly snapshot basis, the period selected is the

half-hourly interval corresponding to the maximum of the average NEM operational “ex VRE”

demand20 across all 10% POE simulations. Most daily outputs represent a half-hourly snapshot,

reported on this basis.

Outputs prescribed under clause 3.7.2(f)(1) and (2) – (4) are based on AEMO peak demand

forecasts and corresponding assumptions, and are not utilised by modelling.

Outputs prescribed under 3.7.2(f)(1A) are based on the demand traces used in the MT PASA

reliability run, adjusted to remove all non-scheduled generation.

Output requirements under clauses 3.7.2(f)(1), (1A), (2) and (4) are supplied in the three-hourly

report, and output under clause 3.7.2(f)(3) can be derived from other information provided, as

explained in Appendix B.

Outputs (5), (5A) and (5B) are also supplied in the three-hourly report as the aggregate value of

participant submitted availabilities, and in the DUIDAVAILABILITY report which shows the DUID

level submitted availabilitities for the next 36 months.

On MT PASA system change implementation21, Outputs (5C) are supplied in the MT PASA

reliability run.

18 The latest report can be found at http://www.aemo.com.au/Electricity/National-Electricity-Market-NEM/Security-and-

reliability/Congestion-information/Network-status-and-capability 19 This report provides a list of the constraint equations for outages that are binding in any of the scenarios. The terms on the the left-

hand side (affected generators and interconnectors) are shown and the constraint set the constraint equation belongs to is

indicated. This then ties back to a description of the outage and NOS. 20 Calculated as the maximum of 48 half hourly average “ex VRE” demands. Average is taken across all 10% POE model runs e.g. 5

historical reference years x 100 iterations = 500 simulations. 21 Expected December 2020.

http://www.aemo.com.au/Electricity/National-Electricity-Market-NEM/Security-and-reliability/Congestion-information/Network-status-and-capability

http://www.aemo.com.au/Electricity/National-Electricity-Market-NEM/Security-and-reliability/Congestion-information/Network-status-and-capability


Doc Ref: 42 of 40

APPENDIX A. MT PASA PROCESS ARCHITECTURE

Figure 2 MT PASA Data Flows

PLEXOSSolution

AggregatorSolution Loader

Case Loader

WARE

Half Hourlyaggregated data

Daily aggregated data

Monthly agg. data (1 table)

Annual agg. data (2 tables) Visualisations

WARE(only keep last 2 runs)

MMSDM

Raw PLEXOS data (compressed)

Azure Platform

Cloud

Participants

NemReports

The MT PASA process operates as follows:

1. The valid Registered Participant bids are loaded into tables in the central Market Management

System (MMS) Database. Bid acknowledgements are returned to Registered participants.

2. All relevant input data is consolidated by the MT PASA Case Loader for loading into the

Reliability and LOLP models. This includes information from participant bids, network limits and

outages, generator parameters, hydro modelling information and model configuration details.

3. The MT PASA Case Loader populates the input models for the Reliability and LOLP runs and

activates the modelling simulations in Azure.

4. The MT PASA Solution Aggregator then aggregates the modelling results which are merged

into a file for transfer out of Azure into the Solution Loader.

5. The Solution Loader loads the file into output tables in the NEM database (WARE).

6. The MT PASA NEM report file is then created from the input information and solution

information.

7. The new MT PASA files are reformatted according to the MMS Data Model (MMSDM) and sent

to each Registered Participant.

8. The visualisations are created from the solution tables, and can be accessed via

https://portal.prod.nemnet.net.au/.

https://portal.prod.nemnet.net.au/


Doc Ref: 42 of 40

APPENDIX B. MEDIUM TERM DEMAND FORECASTING PROCESS

MT PASA modelling is based on operational demand forecasts. Figure 3 gives a pictorial definition

of this demand. Participant bids are received on as “as generated” basis, while demand forecasts

used in MT PASA are on a “sent-out” basis. The difference between the two is the auxiliary load –

the station load that supports the operation of the power station.

The estimated auxiliary load is automatically calculated during the modelling as a fixed percentage

of “as generated power”. The generator auxillary information supplied to the model is based on

AEMO’s latest modelling assumptions22 which are published on the AEMO website. The overall

auxiliary load is therefore dependent on the particular dispatch outcome in each simulation as all

generator types have varying levels of auxiliary load.

Figure 3 AEMO Operational Demand Diagram

B.1 Reliability Run Demand Traces

The methodology for creating “as sent out” half hourly demand trace inputs for modelling is

covered below:

• Representative traces are obtained using at least eight years of historical data.

• Future liquefied natural gas (LNG) export demand is assumed to have a flat profile across the

year and is added to the future Queensland demand traces.

• Projections of future levels of annual underlying energy consumption and maximum demand in

each region are obtained from the most recent published AEMO demand forecasts23.

• Derived operational traces (with rooftop PV added) are “grown” to represent future energy

consumption and maximum demand.

• Forecast rooftop PV is subtracted from the grown trace and retained for separate reporting.

• The assumed impact of behind-the-meter battery storage is also incorporated.

22 The latest information on AEMO’s modelling of generator auxiliary load can be found at https://aemo.com.au/energy-

systems/electricity/national-electricity-market-nem/nem-forecasting-and-planning/forecasting-and-reliability/nem-electricity-

statement-of-opportunities-esoo 23 Available at http://forecasting.aemo.com.au/






Doc Ref: 42 of 40

Although not used in the model, “as-generated” half-hourly demand traces are also created for

the purpose of calculating the required output properties related to demand discussed in section

B.3 below. These traces incorporate assumptions around the annual energy and seasonal

maximum demand contributions from auxiliary load which are based on market simulations24.

B.2 Loss of Load Probability Run Demand Traces

The LOLP run uses abstract operational demand and VRE generation traces that assume high

demand and low VRE generation weather conditions on every day. The abstract traces for each

region are developed as follows:

• For each historical reference year (e.g. 2014/15), take the forecast 10% POE operational “sent

out” demand trace (the same one used for the Reliability Run).

• Determine the regional total of VRE generation in the same reference year, by aggregating the

individual VRE generation traces, taking into account the size/timings of committed new

entrants.

• Subtract total regional VRE generation from demand for that particular reference year to

determine a regional “ex VRE” demand trace.

• For each month/subset of a month25 and day-of-week26 type, find the maximum half-hour

operational “ex VRE” demand value across the historical reference years and record the date

(day and year).

Table 5 Example: Maximum dates and time for Ex VRE Demand in February

Date & Time of

Maximum Ex

VRE demand

for Month

Day of

Week

Historical

Reference

Year

Operational

Demand (MW)

VRE Demand (MW) Ex VRE

Demand

(MW)

19/02/2018 17:00 Monday 1213 3,221 395 2,826

06/02/2018 16:00 Tuesday 910 3,311 555 2,756

07/02/2018 17:00 Wednesday 1617 3,350 276 3,074

08/02/2018 18:00 Thursday 1617 3,197 227 2,971

23/02/2018 17:00 Friday 1112 3,191 321 2,870

24/02/2018 18:00 Saturday 1112 3,119 218 2,902

25/02/2018 17:00 Sunday 1415 3,120 180 2,940

• For each date selected above, record the level of operational “ex VRE” demand, and VRE

generation availability for each VRE generator in each of the 48 half hours within the day, from

the corresponding reference year forecast traces.

• Construct the abstract operational “sent out” demand and individual VRE generation traces

repeating values for each day-of-week type (Monday to Sunday) in the month.

B.3 MT PASA Daily maximum and minimum Demand Values

Under clauses 3.7.2(f)(1) to (3), AEMO is required to prepare and publish the following in respect of

each day covered by the MT PASA:

24 Also available at http://forecasting.aemo.com.au/ 25 Smaller time periods may be used to account for holidays e.g. Christmas, and to better represent months where the early weeks are

demonstrably different than later months based on historical demand patterns. 26 Each day of the week is considered separately – i.e. all Mondays are considered together, then all Tuesdays, and so on.



Doc Ref: 42 of 40

(1) forecasts of the 10% probability of exceedance daily peak load, forecasts of the most

probable daily peak load and forecasts of the time of the peak, on the basis of past trends,

day type and special events, including all forecast scheduled load and other load except for

pumped storage loads;

(1A) the maximum and minimum values of the forecasts of the 10% probability of exceedence

peak load and the forecasts of the most probable peak load, prepared by AEMO in

accordance with paragraph (c)(1);

(2) the aggregated MW allowance (if any) to be made by AEMO for generation from non-

scheduled generating systems in each of the forecasts of the 10% probability of exceedance

peak load and most probable peak load referred to in subparagraph (1);

(3) in respect of each of the forecasts of the 10% probability of exceedance peak load and most

probable peak load referred to in subparagraph (1), a value that is the sum of that forecast

and the relevant aggregated MW allowance referred to in subparagraph (2).

All the modelling in MT PASA is on an operational basis and therefore includes the contribution

from large non-scheduled generation. For the purpose of meeting the requirement in 3.7.2(f),

AEMO calculates the total regional generation from these large non-scheduled generators in each

reference year and subtracts that from the operational as-generated (OPGEN) load traces and

targets.

For the purpose of the daily peak demand values, scheduled loads are assumed to be off at time

of peak if storage based, and considered on if large industrial loads. The possible reduction in

demand from large industrial loads during high price events, including wholesale demand

response, is captured in AEMO’s demand side participation forecast.

Figure 4 shows various measures of demand that are required to be published under the Rules or

are used in the MT PASA modelling and how they relate to each other.

Figure 4 Native demand components27

The abstract demand traces used for the LOLP run are not directly comparable to 10% POE daily

demand met by scheduled and semi-scheduled generators due to a different treatment of VRE.


Doc Ref: 42 of 40

LOLP traces consider output of large non-scheduled generation (i.e. not including small non-

scheduled generators) at times of high “ex VRE” demand.

The components for 3.7.2(f)(1)-(3) are produced in three steps:

• Step 1 – Calculate aggregated MW allowance for non-scheduled generation (3.7.2(f)(2)) by

adding the contribution to peak of large non-scheduled generation to the small non-scheduled

generation values published in AEMO’s latest demand and energy forecasts.

• Step 2 – Derive regional daily peak native demand profiles (3.7.2(f)(3)) using the latest forecasts

of summer and winter peak demand as the basis (reported in three-hourly report) and

statistical analysis of historical weekly demand levels relative to these seasonal peaks and a

similar assessment of peak demand of each weekday relative to the weekly peak demands.

• Step 3 – Derive regional native daily peak demand less non-scheduled generation profiles for

MT PASA (3.7.2(f)(1)) by subtracting the components from Step 1 from Step 2.

These are explained in more detail in the following pages.

Figure 5 Method for developing reported MT PASA daily demand forecasts

Step 1

Non-scheduled demand represents the demand met by both small and large non-scheduled

generation. The small non-scheduled demand is supplied through AEMO’s latest forecasts as an

annual summer and winter figure. The large non-scheduled demand is derived by determining the

average generation from large non-scheduled generation across the top 10 hours in each

reference year. The total non-scheduled generation is calculated as the sum of the small and large

non-scheduled components. This value is published in the three-hourly report and meets the

requirements under Clause 3.7.2(f)(2).

Step 2

The weekly factor profile represents a normalised set of factors (i.e. one factor for each week in the

year) determined by taking the ratios of actual maximum weekly demand to the seasonal demand

published in AEMO’s forecasts for the given historical year. The normalised set of factors are

derived taking historical demand and temperature data into consideration. Refer Figure 6 below.

Note that AEMO uses historical data for the past ten years (if available and relevant) for these

steps.


Doc Ref: 42 of 40

Figure 6 Development of weekly factor profile

The weekday factor profile represents the ratios of daily maximum demand to the maximum

demand of each week in a year. Weekday factors are derived taking historical daily peak demand

data as well as regional public holidays for the past 10 years into consideration. The weekday

factors are used consistently across all weeks of the forecast period when MT PASA demand

forecasts are produced and are derived from operational demand data (that exclude large

wind/solar non-scheduled generation, but which contribution on average on a weekly basis would

see a minimal variation).

Figure 7 Development of weekday factor profile

Step 3

Step 3 consists of deriving the regional 10% POE and 50% POE daily peak native demand less non-

scheduled generation profiles by subtracting non-scheduled demand (step 1) from the daily native

demands calculated in step 2. This meets the requirements of Clause 3.7.2(f)(1).

It should be noted that the demand values published to meet 3.7.2(f)(1), (2) and (4) are not directly

reflective of inputs, that is demand traces, that are used in the reliability run, but are prepared to

meet the requisite rules obligations.

Calculation of maximum and minimum daily peak loads

In contrast, the values published in clause 3.7.2(f)(1A) are almost identical to the range of daily

maximum demands considered across the traces for use in the reliability run, with minor

differences being:

a) The published values assume an expected annual auxiliary load and an auxiliary load at time of

peak to convert from sent-out to as-generated, for better comparison with demand published

by AEMO after each trading period, whereas the demand inputs to the reliability run are sent-

out.

b) The published values are net of all non-scheduled generation based on the assumed profiles

of large non-scheduled generation within each region in each reference year, whereas in the

reliability run large non-scheduled generation (and associated demand) is modelled explicitly.


Doc Ref: 42 of 40

The following published demand values are calculated based on the load traces across each

reference year used in the reliability run, but adjusted as outlined above:

• DEMAND10MAX – calculated as the maximum daily demand across all 10% POE traces.

• DEMAND10MIN – calculated as the minimum daily demand across all 10% POE traces.

• DEMAND50MAX – calculated as the maximum daily demand across all 50% POE traces.

• DEMAND50MIN – calculated as the minimum daily demand across all 50% POE traces.

Figure 8 shows an example of the calculation of the above properties for a single day. This shows

that the DEMAND10MAX and DEMAND10MIN values represent the range of daily maximum

demands considered in the 10% POE MT PASA simulations on that day. Similarly the

DEMAND50MAX and DEMAND50MIN values show the range of daily maximum demands

considered in the 50% POE simulations.

Figure 8 Example of daily demand calculations

MT PASA Weekly Energy

The most probable weekly energy requirement is specified in Clause 3.7.2(f)(4). It is calculated from

the historical reference year half-hourly demand traces described above which excludes the

contribution from non-scheduled generation.

For each demand trace, the weekly energy is calculated as the sum of the half-hourly energy in the

week divided by two28. The average weekly energy across the traces is reported.

28 Division by two is needed as AEMO is summing half-hourly demand values.


Doc Ref: 42 of 40

APPENDIX C: PAIN SHARING

The pain sharing principle of the NEM states that load shedding should be spread pro rata

throughout interconnected regions when this would not increase total load shedding. This is to

avoid unfairly penalising one region for a supply deficit spread through several interconnected

regions.

Specifically, the Equitable Load Shedding Arrangement29 states “as far as practicable, any

reductions, from load shedding as requested by AEMO and/or mandatory restrictions, in each

region must occur in proportion to the aggregate notional demand of the effective connection

points in that region, until the remaining demand can be met, such that the power system remains

or returns (as appropriate) initially to a satisfactory operating state.”

It is open to interpretation whether the pain sharing principles should apply over the annual

period, or be more literally applied to each half-hour period where USE may be projected,

irrespective of previous incidents. One may argue that, for planning purposes, pain sharing should

aim to equalise USE across all NEM regions over the year, taking account of localised USE events

that have already occurred. This would be consistent with implementation of the reliability

standard, using pain sharing to keep load shedding in all regions to less than the level defined in

the reliability standard where possible.

Irrespective of the interpretation of the principle, the EY Report on MT PASA stated that pain

sharing is problematic in models, since shifting USE between regions will almost inevitably change

interconnector losses, generally increasing the total quantity of USE. Since the purpose of MT

PASA is to accurately assess USE, EY recommended that pain sharing be considered a non-core

component of MT PASA design.

AEMO considers that the interests of the markets are best served by providing an accurate

assessment of USE in any region, where shortfall occurs to encourage efficient locational

investment signals.

Application of pain sharing to MT PASA modelling results has the potential to obscure the true

state of supply issues in a region and thus will not be incorporated into the reliability assessments.

29 https://www.aemc.gov.au/sites/default/files/content//Guidelines-for-Management-of-Electricity-Supply-Shortfall-Events.PDF

https://www.aemc.gov.au/sites/default/files/content/Guidelines-for-Management-of-Electricity-Supply-Shortfall-Events.PDF


Doc Ref: 42 of 40

APPENDIX D: CALCULATION OF TRANSFER LIMITS

Interconnector transfer capabilities in the presence of outages are calculated by examining the

results of the MT PASA Reliability Runs according to the following process:

• Obtain the static import and export rating for each interconnector.

• Examine each binding constraint that has the interconnector term on the LHS.

• Move all non-interconnector terms to RHS and calculate RHS value based on dispatch

outcomes.

• Divide the constraints RHS value by the coefficient of the interconnector term on the LHS.

• Positive values refer to an export limit, negative values are imports.

• Set the interconnector limit for a given Monte-Carlo sample equal to the minimum value from

all relevant constraints.

To assess whether interconnector flow is binding for import or export, the following logic is used:

1. Examine list of constraints before they are put into simulation.

2. Flag a group of those constraints as 'Interconnector export limiting' (defined as a constraint

with an interconnector term on LHS and a positive interconnector term factor) -> do this for

each interconnector ID.

3. Flag a group of those constraints as 'Interconnector import limiting' (defined as a constraint

with an interconnector term on LHS and a negative interconnector term factor) -> do this for

each interconnector ID.

For each flagged group of constraints, perform the following aggregation logic for the

probabilities:

PROBABILITYOFBINDINGEXPORT = [Count of all iterations in a specific demand POE level and

time period that have a constraint with both Constraint.HoursBinding>0 and is flagged as

'Interconnector export limiting' for that interconnector id] / [Total number of Iterations]

PROBABILITYOFBINDINGIMPORT = [Count of all iterations in a specific demand POE level and

time period that have a constraint with both Constraint.HoursBinding>0 and is flagged as

'Interconnector import limiting' for that interconnector id] / [Total number of Iterations]


Doc Ref: 42 of 40

APPENDIX E: GRAPHICAL OUTPUTS

The following charts represent outputs that will be available on the AEMO website following each

MT PASA run. They are based on “mock data” and do not represent real modelling outcomes. The

charts in this Appendix are interpretative only.

Figure 9 shows the output from the Reliability Run that indicates whether the reliability standard

can be met in each region for each year of the reliability assessment. The red line indicates the

reliability standard, so any bars that exceed the reliability standard indicate a low reserve condition

exists.

Figure 9 Assessment of Reliability Standard

Figure 10 shows the distribution of unserved energy (USE) across a year and is intended to give

information on the range of USE outcomes observed in each simulation run conducted for

different demand POE levels. The chart indicates that approximately 40% of simulation runs under

10% POE conditions showed the reliability standard was breached, while less than 10% of

simulation runs at the 50% POE level reported a breach of the reliability standard.


Doc Ref: 42 of 40

Figure 10 Annual distribution of Unserved energy (User to select region and year)

Figure 11 shows the distribution of the size of USE events seen in each month in boxplot format.

Only those periods where USE was greater than zero are shown on the plot. The boxes represent

the 25th to 75th percentiles of USE (the median line is in the middle of the box) when comparing

the total monthly USE for each simulation run. The whiskers show the minimum and maximum

values observed across the simulations.

Figure 11 Size of Unserved Energy events by month (User to select POE demand level, region and year)

Figure 12 gives more detailed insight into the USE observed through modelling outcomes by

considering the frequency of events as well as the expected size of the USE events. The chart

shows the 10% POE condition’s USE events are larger in size and more frequent than those of the

50% POE condition.


Doc Ref: 42 of 40

Figure 12 Severity and Frequency of Unserved Energy (User to select region and year)

Figure 13 shows the average interconnector capacity limits (averaged across the Monte-Carlo

simulations) in the presence of network outages as well as a half-hourly snapshot of flow on the

interconnector.

Figure 13 Interconnector flow limits (User to select interconnector)

Figure 14 shows the output from the LOLP run. The grey area shows the scheduled generation

availability according to MT PASA bids. The black line shows the operational demand trace

calculated for the LOLP run with the associated VRE generation (orange area). The top line

represents the total available nameplate capacity (both scheduled and VRE).


Doc Ref: 42 of 40

The Daily LOLP index shown at the bottom of the chart indicates the periods at risk of loss of load

under extreme weather conditions. Periods of relatively high LOLP should be avoided if possible

when scheduling maintenance. The LOLP is colour coded according to the extent of USE expected

in that half hour with the highest loss of load in each day. Red indicates that the magnitude is high

(greater than 400MW), orange that the magnitude is moderate (between 150 MW and 400 MW)

and yellow that the magnitude is low (less than 150 MW).

Figure 14 Supply demand breakdown and maintenance period overview from LOLP run (User to select

region and year)

Figure 15 shows the expected monthly USE for a given year and region. The figure below will be

updated to reflect the final format.

Figure 15 Monthly expected unserved energy (User to select region and year)


Doc Ref: 42 of 40

APPENDIX F: MT PASA OUTPUT TABLES

COLUMN_NAME DATA TYPE DESCRIPTION

MTPASA_CONSTRAINTRESULT

RUN_DATETIME Date processing of the run begins

RUN_NO Unique run id

RUNTYPE Type of run. Always RELIABILITY

DEMAND_POE_TYPE Demand POE type used. Values are POE10

DAY Day this result is for

CONSTRAINTID The unique identifier for the constraint. Only binding

or violating constraints are reported

EFFECTIVEDATE The effective date of the constraint used

VERSIONNO The version of the constraint used

PERIODID Half hourly period reported, selected as period of

maximum NEM operational “ex VRE” demand

(calculated as maximum of “ex VRE” demands,

averaged over reference years and iterations)

PROBABILITYOFBINDING Snapshot – half hourly (NEM Max) Proportion of a constraint binding across iterations

and reference years

PROBABILITYOFVIOLATION Snapshot – half hourly (NEM Max) Proportion of a constraint violating across iterations

and reference years

CONSTRAINTVIOLATION90 Snapshot – half hourly (NEM Max) The 90% percentile violation degree for this

constraint, across iterations and reference years

(MW)


constraint, across iterations and reference years (MW


constraint, across iterations and reference years

(MW)

LASTCHANGED Date the report was created

MTPASA_CONSTRAINTSUMMARY






CONSTRAINTID The unique identifier for the constraint. Only binding

or violating constraints are reported

EFFECTIVEDATE The effective date of the constraint used

VERSIONNO The version of the constraint used

AGGREGATION_PERIOD Snapshot – half hourly

peak/shoulder/off-peak

Period data is aggregated over. Values are PEAK,

SHOULDER, OFFPEAK or PERIOD

CONSTRAINTHOURSBINDING Snapshot – half hourly

peak/shoulder/off-peak

Constraint hours binding for period


Doc Ref: 42 of 40



MTPASA_INTERCONNECTORRESULT






INTERCONNECTORID The unique identifier for the interconnector

PERIODID Half hourly period reported, selected as period of

maximum NEM “ex VRE” demand (calculated as

maximum of “ex VRE” demands, averaged reference

years and iterations)

FLOW90 Snapshot – half hourly (NEM Max) The 90% percentile for flows across iterations and

reference years. Positive values indicate exporting,

negative values indicate importing (MW)







PROBABILITYOFBINDINGEXPORT Snapshot – half hourly (NEM Max) Proportion of iterations and reference years with

interconnector constrained when exporting

PROBABILITYOFBINDINGIMPORT Snapshot – half hourly (NEM Max) Proportion of iterations and reference years with

interconnector constrained when importing

CALCULATEDEXPORTLIMIT Snapshot – half hourly (NEM Max) Calculated Interconnector limit of exporting energy

on the basis of invoked constraints and static

interconnector export limit, averaged across

iterations and reference years

CALCULATEDIMPORTLIMIT Snapshot – half hourly (NEM Max) Calculated Interconnector limit of importing energy

on the basis of invoked constraints and static

interconnector import limit, averaged across

iterations and reference years


MTPASA_LOLPRESULT



RUNTYPE Type of run. Always LOLP


REGIONID The unique region identifier

WORST_INTERVAL_PERIODID Snapshot – half hourly (worst of

day)

The half hourly interval period with the highest LOLP,

or highest region demand net of VRE generation if

LOLP = 0 for all intervals (1..48)

WORST_INTERVAL_DEMAND Snapshot – half hourly (worst of

day)

The LOLP half hourly operational as-generated

demand for the worst interval in this region (MW)


Doc Ref: 42 of 40


WORST_INTERVAL_INTGEN Snapshot – half hourly (worst of

day)

The half hourly aggregate VRE generation for the

interval period with the worst LOLP in this region

(MW)

WORST_INTERVAL_DSP Snapshot – half hourly (worst of

day)

The half hourly aggregate demand side participation

for the interval period with the worst LOLP in this

region (MW)

LOSSOFLOADPROBABILITY Snapshot – half hourly (worst of

day)

Loss of Load Probability for day reported

LOSSOFLOADMAGNITUDE Snapshot – half hourly (worst of

day)

Loss of Load Magnitude for day reported. Values are

LOW, MEDIUM, HIGH


MTPASA_REGIONRESULT







PERIODID Snapshot – half hourly (NEM Max) Half hourly period reported, selected as period of

maximum NEM “ex VRE” demand (calculated as

maximum of “ex VRE” demands, averaged reference

years and iterations)

DEMAND Snapshot – half hourly (NEM Max) OPGEN Demand value from selected half hourly

interval (MW). This value includes contribution from

large non-scheduled generation so may be higher

than the demand values published in the

REGION_AVAILABILITY table.

AGGREGATEINSTALLEDCAPACITY Snapshot – half hourly (NEM Max) The total rated capacity of all active generation (MW)

NUMBEROFITERATIONS Snapshot – half hourly (NEM Max) Total number of iterations and reference years

performed

USE_NUMBEROFITERATIONS Snapshot – half hourly (NEM Max) Number of iterations and reference years showing

USE

USE_AVERAGE Snapshot – half hourly (NEM Max) Average USE across all iterations and reference years

(MW)

USE_EVENT_AVERAGE Snapshot – half hourly (NEM Max) Average USE event size across all iterations and

reference years (MW)

USE_MAX Snapshot – half hourly (NEM Max) Maximum USE across all iterations and reference

years (MW)

USE_MIN Snapshot – half hourly (NEM Max) Minimum USE across all iterations and reference

years (MW)

USE_MEDIAN Snapshot – half hourly (NEM Max) Median USE across all iterations and reference years

(MW)

USE_LOWERQUARTILE Snapshot – half hourly (NEM Max) Lower quartile USE across all iterations and reference

years (MW)

USE_UPPERQUARTILE Snapshot – half hourly (NEM Max) Upper quartile daily USE across all iterations and

reference years (MW)


Doc Ref: 42 of 40


TOTALSCHEDULEDGEN90 Snapshot – half hourly (NEM Max) The 90% percentile for scheduled generation across

iterations and reference years (MW)





TOTALINTERMITTENTGEN90 Snapshot – half hourly (NEM Max) The 90% percentile for VRE generation across all






TOTALSEMISCHEDULEDGEN90 Snapshot – half hourly (NEM Max) The 90% percentile for semi-scheduled generation

across all iterations and reference years (MW)





DEMANDSIDEPARTICIPATION90 Snapshot – half hourly (NEM Max) The 90% percentile for demand side participation

across all iterations and half hours (MW)





TOTALAVAILABLEGEN90 Snapshot – half hourly (NEM Max) The 90% percentile for total Scheduled availability






TOTALAVAILABLEGENMIN Snapshot – half hourly (NEM Max) The minimum for total Scheduled availability across

all iterations and half hours (MW)

TOTALAVAILABLEGENMAX Snapshot – half hourly (NEM Max) The maximum for total Scheduled availability across

all iterations and half hours (MW)


MTPASA_REGIONSUMMARY




DEMAND_POE_TYPE Demand POE type used. Values are POE10 or POE50

AGGREGATION_PERIOD Period data is aggregated over. Values are YEAR,

MONTH

PERIOD_ENDING Date time of day at end of interval (which may be

over a year, a month)



Doc Ref: 42 of 40


NATIVEDEMAND Average monthly/annual iteration

totals

Native demand from AEMO forecast, pro-rated for

horizon year specified in PERIOD_ENDING (MWh)

USE_PERCENTILE10 Percentiles assessed over iteration

totals for either month or year

USE period amount at the 10% percentile of iterations

and reference years (MWh)



USE period amount at the 20% percentile of

iterations and reference years (MWh)



















USE period amount at the 70% percentile of iterations

and reference years (MWh)













USE_AVERAGE Average monthly/annual iteration

totals

Average period USE across iterations and reference

years (MWh)

WEIGHT Fixed value Weighting use for aggregating POE Demand Level.

0.392 (50 POE) or 0.304 (10 POE)

USE_WEIGHTED_AVG Regional Weighted Average USE

(Percent)

((USE_AVERAGE_POE10 / NATIVE_DEMAND_POE_10

* WEIGHT_POE_10) + (USE_AVERAGE_POE50 /

NATIVE_DEMAND_POE_50 * WEIGHT_POE_50))*100

LRC LRC reporting for region LRC Condition reported (Value=1) if

USE_WEIGHTED_AVG >= 0.002% otherwise no LRC

(Value=0)

NUMBEROFITERATIONS Value by month/year Total number of iterations and reference years

performed

USE_NUMBEROFITERATIONS Value by month/year Number of iterations and reference years showing

USE

USE_EVENT_UPPERQUARTILE Assessed over iteration totals for

either month or year

Upper quartile USE event size across all half hourly

intervals and iterations and reference years that have

USE>0 (MW)

USE_EVENT_LOWERQUARTILE Assessed over iteration totals for


Lower quartile USE event size across all half hourly


USE>0 (MW)

USE_EVENT_MAX Assessed over iteration totals for


Max quartile USE event size across all half hourly


USE>0 (MW)

USE_EVENT_MIN Assessed over iteration totals for


Min quartile USE event size across all half hourly


USE>0 (MW)


Doc Ref: 42 of 40


USE_EVENT_MEDIAN Assessed over iteration totals for


Median quartile USE event size across all half hourly


USE>0 (MW)


MTPASA _REGIONITERATION




DEMAND_POE_TYPE Demand POE type used. Values are POE10 or POE50

AGGREGATION_PERIOD Period data is aggregated over. Values are YEAR,

MONTH

PERIOD_ENDING Date time of day at end of interval (which may be

over a year, a month)


USE_ITERATION_ID - Iteration ID, only produced for iterations showing

unserved energy>0

USE_ITERATION_EVENT_NUMBER Value by month/year Number of half hours showing unserved energy over

year, for iteration

USE_ITERATION_EVENT_AVERAGE Assessed over iteration totals for


Average unserved energy event size for iteration over

year (MW)


MTPASA_REGIONAVAILABILITY – THREE-HOURLY REPORT

PUBLISH_DATETIME Date Time the report was published.

DAY Date on which the aggregation applies.

REGIONID NEM Region

PASAAVAILABILITY_SCHEDULED Regional aggregation of bid values Aggregate of the offered PASA Availability for all

Scheduled generators in this region.

LATEST_OFFER_DATETIME Date Time of the latest offer used in the aggregation

for this region and date.

ENERGYUNCONSTRAINEDCAPACIT

Y

Region energy unconstrained MW capacity

ENERGYCONSTRAINEDCAPACITY Region energy constrained MW capacity

NONSCHEDULEDGENERATION Daily Peak Allowance made for small non-scheduled generation

in the demand forecast (MW).

DEMAND10 Daily Peak 10% POE peak demand, as-generated, excluding

non-scheduled generation and scheduled load.

DEMAND50 Daily Peak Most probable peak demand, as-generated,

excluding non-scheduled generation and scheduled

load.

DEMAND10MAX Daily Peak Maximum of the scheduled demand peaks that occur

in the 10% POE traces (MW), excluding scheduled

load and non-scheduled generation.


Doc Ref: 42 of 40


DEMAND10MIN Daily Peak Minimum of the scheduled demand peaks that occur



DEMAND50MAX Daily Peak Maximum of the scheduled demand peaks that occur



DEMAND50MIN Daily Peak Minimum of the scheduled demand peaks that occur



ENERGYREQDEMAND10 Weekly Total Weekly Energy (Operational as-generated, excluding

scheduled load) calculated directly from the half

hourly 10% POE trace (GWh).

ENERGYREQDEMAND50 Weekly Total Weekly Energy (Operational as-generated, excluding

scheduled load) calculated directly from the half

hourly 50% POE trace (GWh).


MTPASA _ DUIDAVAILABILITY – THREE-HOURLY REPORT

PUBLISH_DATETIME Date Time the report was published.

DAY Date on which the aggregation applies.

DUID Unit level DUID

REGIONID Region ID for the DUID

PASAAVAILABILITY PASA availability (MW)

LATEST_OFFER_DATETIME Date Time of the latest offer used for this unit and

date.



Doc Ref: 42 of 40

APPENDIX G: “PLAIN ENGLISH” REPORT ON CONSTRAINTS

AEMO will provide a “plain English” report on constraints that provides further details on

generators impacting by binding constraints.

To access the “plain English report” service:

1. Access via: https://portal.prod.nemnet.net.au/#/signin

2. From menu items: MMS➔Market Info➔View Constraints➔View Constraints

3. Type in constraints in the Constraints ID field, and Submit as per the screenshot below. The

“plain English” report will be displayed on submission.

Figure 16 Example constraints viewer

https://portal.prod.nemnet.net.au/#/signin

medium term pasa process description - aemo

Documents